WO2021110176A1

WO2021110176A1 - Edge system and method for processing data operation request

Info

Publication number: WO2021110176A1
Application number: PCT/CN2020/134345
Authority: WO
Inventors: 殷薇
Original assignee: 华为技术有限公司
Priority date: 2019-12-06
Filing date: 2020-12-07
Publication date: 2021-06-10
Also published as: US20220303718A1; US11895554B2; CN112929459B; EP4057585A1; CN112929459A; EP4057585A4

Abstract

The present disclosure relates to an edge system. The edge system comprises a plurality of first-level circle layers and a second-level circle layer, a geographical area represented by the second-level circle layer comprises geographical areas represented by each of the first-level circle layers, that is, a plurality of first-level circle layers are deployed in the second-level circle layer. Each of the first-level circle layers has a plurality of sites, the plurality of sites comprise an initial site and are included in the second-level circle layer, and the initial site processes a data operation request sent by a client in the first-level circle layer and the second-level circle layer comprising the initial site. The edge system enables more secure and efficient data processing.

Description

Edge system and processing method of data operation request

Technical field

This application relates to the field of computer technology, in particular to an edge system and a method for processing data operation requests executed by the edge system.

Background technique

With the advancement of mobile communication technology, there are more and more low-latency and computationally intensive services, such as virtual reality (VR), augmented reality (AR), and Internet of Vehicles. In order to reduce the computing power requirements of such services on the client and provide low-latency or even ultra-low-latency services to the client, cloud service providers deploy sites near the edge of users. The geographic locations of these sites may be highly discrete, and the server clusters in each site are generally not large, but the number of sites can be relatively large, and the networks between sites are often not guaranteed by dedicated lines.

In order to manage these highly discrete sites, one way is to use a traditional centralized management architecture, as shown in Figure 1. In this architecture, after site installation and deployment are started, site information needs to be registered with the central management site, and the site is required to periodically confirm the connection survival with the central management site. The central management site needs to deploy database services to manage the host resources, tenant resources and other information of each site. Since this information is stored in the central management site, the issuance, management, and query of resources required by the cloud services running on the site all need to manipulate this information.

The disadvantage of the centralized management architecture is that the huge fan-out center-and-spoke topology will have problems such as limited site management scale and unsatisfied low-latency user requests. With the increase in the number of sites and the further expansion of the geographical dispersion of sites, the centralized management architecture will become a management bottleneck.

Summary of the invention

This application provides an edge system, which implements mobile data following through a multi-circle-layer site design, and improves the processing efficiency of data operation requests and data security.

The first aspect provides an edge system, the edge system includes a plurality of first-level circle layers and second-level circle layers, and the geographic area represented by the second-level circle layer includes each of the first-level circle layers. Geographical area. Each first-level circle layer includes a plurality of sites, and each site is used to process data operation requests sent by a client, and an initial site among the plurality of sites is also included in the second-level circle layer. The initial site is stored in a certain first-level circle layer and second-level circle layer at the same time, and is used to process the data operation in the first-level circle layer and the second circle layer including the initial station request.

The edge system provided above is designed with multi-level circle sites, and there are site intersections between the multi-level circle layers. These intersection sites can implement data operations in each circle layer, ensuring that data is in different first-level circles. In the case of moving between layers, data security, reliability, and data access efficiency can still be guaranteed.

In a possible implementation manner, when the data operation request is a write request; the initial site is used to address the target carried in the data operation request in the first-level circle layer including the initial site The target site corresponding to the data, the target data is sent to the target site of the first-level circle including the initial site; the target site corresponding to the target data is addressed in the second-level circle, and all The target data is sent to the target site of the second-level circle. The target site including the first level circle layer of the initial site is used to store the target data. The target site of the second level circle layer is used to store the target data.

In the case of a write request, the initial site writes the target data in both the first-level circle and the second-level circle, so that if the client subsequently accesses the target data in the same first-level circle, it can efficiently access the target data. Obtain target data in the first-level circle site. At the same time, when the client accesses the target data after moving to other first-level circles, since the original first-level circles and the moved first-level circles belong to the same second-level circle, they can pass through the second level. The target data stored in the level circle layer provides services for the client, ensuring the security and reliability of the client's access to the data.

In a possible implementation manner, the target site that includes the first-level circle of the initial site is also used to create or update the target data on other sites that include the first-level circle of the initial site Copy.

In a possible implementation manner, the target site of the second-level circle is also used to create or update a copy of the target data on other sites of the second-level circle.

By realizing the copy of the target data at all levels, the security of the target data is improved.

In a possible implementation manner, when the data operation request is a read request; the initial site is used to address the target carried in the data operation request in the first-level circle layer including the initial site The target site corresponding to the key of the data; the target site of the first-level circle including the initial site, used to determine whether the target data is stored locally in the target site of the first-level circle including the initial site, If the target data is stored locally at the target site including the first level circle layer of the initial site, the target data is sent to the client; if the target data is not stored in the first site including the initial site The target site of the first level circle layer is local, and the target site corresponding to the key of the target data is addressed in the second level circle layer; the target site of the second level circle layer is used to determine the target data storage Send the target data to the client locally at the target site of the second-level circle.

In the case of a read request, as described above, first try to find the target data in the first-level circle that is closer to the client. If it cannot be found, search for the target data in the second-level circle. The efficiency and safety of data operation are both considered.

In a possible implementation manner, the target site of the second-level circle is also used to send the target data to the target site of the first-level circle that includes the initial site.

Send target data to the target site of the first-level circle that includes the initial site, so that when the target data is subsequently accessed in the first-level circle that includes the initial site, it can be directly included in the initial site. It is found in the first-level circle, which improves work efficiency.

The second aspect provides a method for processing a data operation request. The data operation request is a write request. The processing method for the data operation request is applied to an edge system. The edge system includes a plurality of first-level circles and a second Level circle layer, the geographic area represented by the second level circle layer includes the geographic area represented by each first level circle layer, each first level circle layer includes multiple sites, and each site is used to process the data sent by the client A data operation request, including an initial site in the plurality of sites, is also included in the second-level circle; the initial site addresses the data operation request carried in the first-level circle that includes the initial site The target site corresponding to the target data of the initial site; the initial site sends the target data to the target site of the first-level circle that includes the initial site; the target site storage of the first-level circle that includes the initial site The target data; the initial site addresses the target site corresponding to the target data in the second-level circle, and sends the target data to the target site of the second-level circle; the second The target sites in the tier circle store the target data.

The actions of addressing the initial site in the first level circle layer including the initial site and the actions of addressing the initial site in the second level circle layer may be performed in parallel.

In a possible implementation manner, the method further includes: the target site of the first-level circle including the initial site creates or updates the target data on other sites of the first-level circle including the initial site a copy of.

In a possible implementation manner, the method further includes: the target site of the second-level circle creates or updates a copy of the target data on other sites of the second-level circle.

A third aspect provides a method for processing a data operation request. The data operation request is a read request. The processing method for the data operation request is applied to an edge system. The edge system includes a plurality of first-level circles and a second The geographic area represented by the second-level circle layer includes the geographic area represented by each first-level circle layer, and each first-level circle layer includes multiple sites, and each site is used to process client sending The data operation request of the plurality of sites including an initial site is also included in the second-level circle; the initial site addresses the data operation request in the first-level circle that includes the initial site The target site corresponding to the key of the carried target data; the target site of the first-level circle including the initial site determines whether the target data is stored locally in the target site of the first-level circle including the initial site; If the target data is stored locally in the target site including the first level circle layer of the initial site, the target data is sent to the client; if the target data is not stored in the first site including the initial site The target site of the first level circle layer is local, and the target site corresponding to the key of the target data is addressed in the second level circle layer; the target site of the second level circle layer determines that the target data is stored in the The target site of the second-level circle layer sends the target data to the client locally.

In a possible implementation manner, the method further includes: the target station of the second-level circle layer sending the target data to the target station of the first-level circle layer including the initial station.

The target station of the second-level circle layer sends the target data to the target station of the first-level circle layer including the initial station and the target station of the second-level circle layer sends the target data to all The actions of the client can be executed in parallel.

The fourth aspect provides a non-transitory readable storage medium. When the instructions stored in the non-transitory readable storage medium are executed by each edge site system, the system executes the aforementioned third aspect and its possible implementation Or the method provided by the fourth aspect and its possible implementations. Program instructions are stored in the storage medium. The storage medium includes, but is not limited to, volatile storage devices, such as random access storage devices, and non-volatile storage devices, such as flash storage devices, hard disk drives (HDD), and solid state drives (SSDs). ).

The fifth aspect provides a computer program product. When the instructions contained in the computer program product are executed by the edge system, the system executes the foregoing third aspect and its possible implementation manners or the fourth aspect and its possible implementation manners. . The computer program product may be a software installation package. In the case of using the method provided in the first aspect and its possible implementation manners, the computer program product may be downloaded and executed on each site of the edge system The instructions included.

Description of the drawings

In order to explain the technical methods of the embodiments of the present application more clearly, the following will briefly introduce the drawings that need to be used in the embodiments.

Figure 1 is a schematic diagram of the structure of the edge system provided by this application;

Figure 2 is a schematic structural diagram of another edge system provided by this application;

Figure 3 is a schematic structural diagram of an edge site provided by this application;

FIG. 4 is a schematic structural diagram of a hash ring of an addressing algorithm provided by this application;

FIG. 5 is a schematic structural diagram of a hash ring of another addressing algorithm provided by this application;

FIG. 6 is a schematic structural diagram of a hash ring of another addressing algorithm provided by this application;

FIG. 7 is a processing flowchart of a data operation request provided by this application;

FIG. 8 is a processing flowchart of another data operation request provided by this application;

Fig. 9 is a schematic structural diagram of a computer provided by this application.

Detailed ways

The technical methods in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

As shown in Figure 2, the present application provides an edge system. The edge system includes multiple sites. The edge system includes multi-level circles. The edge system can have one or more circles of the same level. Each site can belong to multiple circles of different levels at the same time. Each site cannot belong to multiple circles at the same time. Level circle layer. Figure 2 shows only a part of the edge system, and other possible devices, such as the edge gateway of the edge system, are omitted.

Circles, generally speaking, different circles represent different geographic areas. The high-level circles represent a larger geographic area, and the low-level circles represent a smaller geographic area. Different circles at the same level represent different geographical areas at the same level. The geographic area represented by each high-level circle includes at least one low-level circle. For example, the third-level circle represents a country, the second-level circle represents a province, and the first-level circle represents a city. Then, the geographic area represented by the third-level circle layer includes at least one second-level circle layer, and the geographic area represented by each second-level circle layer includes at least one first-level circle layer. The sites in the high-level circle belong to a certain low-level circle included in the geographic area represented by the high-level circle. For example, site 1 in the third-level circle belongs to a certain second-level circle included in the third-level circle, and site 1 also belongs to a certain first-level circle included in the second-level circle. Stations 13 to 19 in FIG. 2 actually belong to a certain first-level circle, and each station included in the third-level circle actually belongs to a certain second and first circle. The stations in each circle layer belong to an addressing domain, such as a hash domain. For example, stations 1 to 6 belong to a first-level addressing domain, and stations 7 to 12 belong to another first-level addressing domain, site 1 , Site 13 to Site 19 belong to a second-level addressing domain, this part will be expanded later.

The higher the level of the circle, the more scattered the physical area of the site deployment, so the delay of providing services to the client is also higher. For example, the delay of the station in the third level circle to respond to the client request is 50 milliseconds, and the second level circle The delay for stations in the layer to respond to client requests is 20 milliseconds, and the delay for stations in the first level circle to respond to client requests is 10 milliseconds. Therefore, the edge system will respond to client requests as much as possible with sites in the lower-level circle that are close to the client.

The client can be a hardware terminal, such as a computer, a mobile phone, or a software module running on the hardware terminal. The client generates data operation requests according to business requirements, such as read requests/write requests, and sends the data operation requests to the edge system to operate on the data stored in the edge system. The data operation request includes information such as the key and operator of the target data. The target data is the operation object of the data operation request, and the key is the information such as the access path and access address of the target data.

Each site includes at least one operating unit, and each operating unit is a physical computer or a virtual machine. The modules included in each site can be distributed to run on each operating unit included in the site. When a site includes multiple operating units, the multiple operating units can store each target data in a distributed manner, and each operating unit stores a part of the target data (a fragment of the target data). The multiple operating units include a metadata server. When a site includes multiple operating units, the site described below is actually the metadata server. The communication network between the stations and the communication network between the station and the client can be a wired communication network or a wireless communication network, for example: the fifth-generation mobile communication technology (5th-Generation, 5G) system, long-term evolution (long term evolution, LTE) system, wireless fidelity (wireless fidelity, WiFi).

As shown in Figure 3, the site includes a data storage module for actually storing data. The data storage module also stores metadata. The metadata records the correspondence between the key of the data and the storage address of the data. Taking object storage as an example in the data storage module, the metadata records the correspondence between the key of the data and the address of the bucket where the data is located.

The station also includes an addressing module, which is used for addressing in the circle layer where the station is located according to the data operation request after the station receives the data operation request. When a site belongs to multiple circle layers, the addressing module includes multiple circle layer addressing modules, and each circle layer addressing module manages an addressing domain. When the low-level circle-level addressing module cannot address the site where the target data is located in this circle-level, the high-level circle-level addressing module continues to address in the high-level circle. Taking site 1 in Figure 2 as an example, site 1 belongs to the first-level circle, the second-level circle, and the third-level circle at the same time. Therefore, after station 1 receives the data operation request, it is first addressed by the first-level circle-level addressing module in the first-level circle. If the station where the target data is located cannot be found in the first-level circle, it will be The second-level circle-level addressing module addresses in the second-level circle. If the site where the target data is still cannot be found in the second-level circle, the third-level circle-level addressing module will be in the third level. Addressing within the circle layer. The specific addressing algorithm adopted by the addressing module will be introduced later.

The site also includes a storage interface processing module, which is used to receive data operation requests, parse the data operation requests, and send the parsed parameters to the required modules.

The site also includes a copy placement strategy configuration module, which is used to configure a copy placement strategy for data requested to be written/modified by a data operation. For example, synchronous processing between replicas in the same circle layer, two copies of three copies in the same circle layer synchronously processing one copy asynchronous processing, and asynchronous processing between copies of the low-level circle layer and the high-level circle layer, etc.

The station also includes an addressing domain configuration module, which is used to create and modify (when a station is added) each addressing domain, that is, to configure the addressing domain module.

Optionally, the site can also run various services. For example, database services, various applications, etc. In one case, when the client accesses the edge system, it may regard the edge system as a storage system, and directly send data operation requests to the edge gateway of the edge system, and the edge gateway sends the data operation request to the storage interface of a certain site for processing Module. In this case, the site does not need to run various services.

In another case, a service is running on the site. Taking the database service as an example, the client regards the edge system as a database service, the client sends the service call request of the database to the edge gateway, and the edge gateway sends the service call request to A database service running on a site. After the database service executes the database operation request, it generates a data operation request corresponding to the database operation request, and sends the data operation request to the storage interface processing module.

In the edge system, the target data of the data operation request may be stored in a site in the lower level circle, or it may be stored in a site in the high level circle. The following describes the addressing algorithm in the edge system (that is, looking for The way of the site where the target data is located).

The data operation request generated by the client is sent to the initial site, and the addressing module in the initial site first addresses the addressing domain of the lowest level circle layer where the initial site is located. Specifically, the addressing algorithm in the addressing domain may adopt flexible routing table (flexible routing table), distributed hash table (distributed hash table) and other methods for addressing. Or the addressing module of each site can directly record the metadata of the data stored in each site in the circle, and the addressing module of the initial site can directly search for the site where the target data is located. This situation is also called the entire network格 (full mesh). In the case of a full grid, since the addressing module needs to record a large amount of metadata, a large amount of storage space is occupied. At the same time, when the target data stored in any site is modified, the metadata of the addressing module of each site needs to be modified synchronously, resulting in relatively large communication consumption. Therefore, the full grid is suitable for smaller circles. Floor. When the scale of the circle is large and the number of sites in the circle is large, the distributed hash table can be used for addressing, or further, the flexible routing table can be used for addressing.

The following uses a distributed hash table as an example to introduce the addressing algorithm in the addressing domain of a circle layer. First, each piece of data stored in the circle layer is represented as a KV (key, value) pair, and the value is the IP address of the site where the data is stored (or other description information of the site). These KV pairs constitute the index table of the data stored in the circle layer. According to specific rules, the index table is split and distributed to each site in the circle layer, so that each site is responsible for maintaining part of the index table. Then, by continuously forwarding data operation requests between the sites, it is constantly approaching the site where the target data is located.

Take Chord algorithm as an example to introduce an implementation of distributed hash table. Chord guarantees consistent hashing by mapping sites and keys to the same space. First, a hash function is used to generate a 2 ^m space (2 ^{m is} not less than the largest NID). The integers in the space are connected end to end to form a ring, which is called a Chord ring. Each point on the Chord ring is called a marker. If a certain site is mapped to a certain identifier, the identifier is called a node. According to the clockwise direction on the Chord ring, the node located in front of a node is called the predecessor node of the node, and the node located behind the node is called the successor node. The first successor node is called the immediate successor node, and the first successor node is called the immediate successor node. After the site is hashed to the Chord ring, the node identifier (NID, node identifier) can be obtained. The NID is obtained by the hash operation from the IP address of the site, and the key identifier (KID) can be obtained after the key is hashed on the Chord ring. ), KID is obtained from the key through hash operation. The NID is placed on the Chord ring clockwise from small to large. The data corresponding to each KID is allocated to the site with NID≥KID for storage. As shown in Figure 4, for the Chord ring with m=6, there are 10 NIDs and 5 KIDs. Among them, the data corresponding to KID 10 is allocated to NID 14 for storage, and the data corresponding to KID 24 and KID 30 are allocated to NID. 32, the data corresponding to KID 38 is allocated to NID 38 for storage, and the data corresponding to KID 54 is allocated to NID 56 for storage.

Each NID maintains a routing table with at most m entries (also called finger table) for addressing. The routing table of a NID contains at least the position of ^{the NID+2 (m-1).} Based on the example shown in FIG. 4, refer to FIG. 5. In FIG. 5, KID is denoted as K, and NID is denoted as N. The left column of the N8 routing table contains locations from N8+1 to N8+32, and the right column is the actual site corresponding to each location. For example, the first row (that is, the correspondence between N8+1 and N14) indicates that the data at the first position after N8 is responsible for N14, and the second row (that is, the correspondence between N8+2 and N14) It means that N14 is responsible for the data in the second position after N8, and the meanings of other rows are similar. Therefore, the routing table of each node contains only a small part of the node's information. The routing table of each node records more locations that are responsible for neighboring successor nodes. For example, N8 records 3 data responsible for N14, while only 1 data responsible for N21 is recorded. When searching for a resource on a node, first determine whether its successor node holds the resource, if not, start the search directly from the farthest point in the routing table of the node to see which item is away from the one holding the data The node is the most recent and sends an addressing request to the node. If not, it means that the node itself has the resource to be found, so iterate.

For example, referring to Figures 5 and 6, looking for the data corresponding to K54 in N8, first, search the N8 routing table for the successor node N14, and find that K54 does not meet the requirements of 54 ∈ (8; 14), then in the N8 routing table Look up the entry that meets the requirements (from far to near, take Figure 5 as an example, that is to look up from the bottom of the routing table), and find that the farthest entry in the routing table N8+32->N42 satisfies 42 ∈ (8 54], it means that N42 is closest to the node holding the data corresponding to K54, then jump to N42 to continue searching (first hop). It can be known from the routing table of N42 that the successor node of N42 is N48, which does not match The requirement of 54∈(42;48] indicates that N48 does not hold the data corresponding to K54, so it starts to look up on the routing table of N42. In the routing table of N42 node, it starts to look up from far to near, and it is found that N42+8-> If N51 satisfies 51 ∈ (42; 54], it means that N51 is closest to the node holding the data corresponding to K54, then jump to the N51 node to continue searching (the second hop). The successor node of the N51 node is N56, which conforms to 54 Ε(51;56], addressing is complete. The data query request of KID 54 is sent to the station corresponding to NID 51 for processing (hop 3).

In the case of newly added sites, the Chord ring will also be refreshed accordingly. The new sites will form new nodes on the Chord ring, and the routing table of each node will be updated accordingly.

If a client originally stored data in the first-level circle (for example, the Wuhan city circle), the first-level circle belongs to a second-level circle (for example, the Hubei province circle). As the client moves to other first-level circles, when the client tries to access the data in the moved first-level circles, the status of each site in the moved first-level circles The addressing module will find that no site can be addressed to store the data in this circle.

In order to avoid this situation and ensure that the mobile data of the client follow, when the client manipulates the data in the lower level, the data will be synchronized to the higher level or even the higher level. In this way, when the client moves to other first-level circles, although the other first-level circles cannot be addressed to a site to store the target data, if the client's original first-level circle and the The moved first-level circle belongs to the same second-level circle, so if the client cannot address the target data in the moved first-level circle, it can be addressed in the second-level circle The target data, and read the target data to the station in the moved first-level circle for subsequent visits. Even if the client’s original first-level circle and the moved first-level circle do not belong to the same second-level circle, the client's original first-level circle and the moved first-level circle belong to the same In a higher-level circle, if the client cannot address the target data in the moved first-level circle, the target data can be addressed in the higher-level circle, and the target The data is read to the station in the first level circle layer after the movement.

In the following, the implementation process of the edge system provided in this application is introduced with the data operation request as the write request and the read request respectively.

When the data operation request is a write request, Figure 7 shows the process of processing the data operation request by the edge system.

S201: The client requests the edge gateway to send a site access request to request address information of the site.

Commonly, the client and the edge gateway communicate through the HTTP/HTTPS protocol, and the client accesses the edge gateway through the domain name of the edge system. In this case, the edge gateway may be a domain name server (domain name service, DNS). The edge gateway selects an initial site for the client based on one or more parameters such as the delay between the client and each site, the workload of each site, and the physical distance between the client and each site.

S202: The edge gateway sends the selected address information of the initial site to the client.

S203, as described above, the client may send a data operation request or a service invocation request (in the case where the client's direct purpose is to invoke a service running on the site) to the initial site.

S204: The initial site obtains the data operation request, and parses the data operation request through the storage interface processing module.

In the case that the client sends a service invocation request in S203, the service running on the initial site generates a data operation request according to the service invocation request.

The data operation request includes the key and the value of the target data. Optionally, the data operation request may also include circle layer information, and the circle layer information may include a default or indicate a certain level of circle layer. among them,

Default: Instruct the site to complete the writing of target data and the creation/update of the copy in the lowest level circle where the site is located according to the preset method in the copy placement strategy configuration module, and complete the writing of target data to the high level circle. Import and copy creation/update to achieve mobile data follow-up. In which advanced circle layers, data follow can be configured.

Specified_layer: Instruct the site to write target data only in the specified level circle. The method of writing the target data can also complete the writing of the target data in the specified level circle according to the preset method in the copy placement strategy configuration module And the creation/update of copies.

In the case of a specified level, it generally means that the data written this time gives up moving data to follow. Therefore, the target data does not need to be written to the advanced layer.

With the circle information as the default, continue to introduce the processing flow of the data operation request.

S205, the lowest-level circle-level addressing module of the initial site, according to the key of the target data, addresses within the lowest-level circle where the initial site is located, and determines the target site corresponding to the key of the target data.

Since the initial site is generally a site closer to the client, not the site responsible for storing the target data specified in the addressing domain, the initial site needs to determine a target based on the key of the target data in the addressing domain where the lowest level circle is located. Site. Continuing to take the algorithm of Figure 4 as an example, assume that the initial site is the site corresponding to NID 8. If the key of the target data is hashed to obtain a KID of 40, then on the hash ring shown in Figure 6, KID 40 should be stored in NID 42 on the corresponding site. Then, the site corresponding to NID 8 needs to send the target data to the site corresponding to NID 42 for storage.

If the initial site happens to be the target site, there is no need to perform S206 after S205.

S206: The initial site sends the key of the target data and the target data to the target site.

After determining the target site, the initial site determines the address information of the target site according to the index table recorded by itself. According to the address information of the target site, the key and target data of the target data are sent to the target site.

S207: The target site stores the target data.

The target site stores the target data, and the target site updates the locally recorded metadata, recording the corresponding relationship between the key of the target data and the storage address of the target data.

S208: The target site creates/updates the replica in the lowest level circle according to a preset method in the replica placement strategy configuration module. If the target data is newly written data, the target site will create a copy of the target data on other sites in the lowest level circle. If the target data is an overwrite of existing data, then the target site will update a copy of the existing data on other sites in the lowest level circle. The specific update method can be synchronous processing of all copies, synchronous processing of partial copies and asynchronous processing of partial copies, and so on.

After the target site determines that S208 is successfully executed, it judges that this writing is successful.

S209: The target site notifies the initial site that the target data is written successfully.

S210: The initial site notifies the client that the target data is successfully written.

If the circle level information carried in the data operation request is a designated level, the processing flow of the data operation request is as described in S205 to S210, and the lowest level circle level in S205 to S210 can be replaced with the designated level circle level.

If the circle layer information carried in the data operation request indicates the default, after S208, the subsequent process needs to be executed.

In the case where the replica placement strategy configuration module indicates that the replicas between different levels of circles are processed asynchronously, the execution of S209 can be performed in parallel with the execution of subsequent steps, that is, the execution of S210 does not have to wait for the completion of the replicas in the advanced circles. When the replica placement strategy configuration module indicates that the replicas between different levels of circles are synchronized, the execution of S210 needs to wait for the replicas in the advanced circles to be processed, that is, it needs to be executed after confirming that the writing is successful in S209 and S215. S210.

In S211, the advanced circle layer addressing module of the initial site addresses the advanced circle layer where the initial site is located according to the key of the target data, and determines other target sites corresponding to the key of the target data in the advanced circle layer.

If in S205, the target site determined by the first-level circle-level addressing module of the initial site, then in S211, the second-level circle-level addressing module of the initial site continues to search for other target sites in the second-level circle.

Generally, other target sites determined in the advanced circle layer are the sites responsible for storing the target data in the addressing domain of the advanced circle layer.

S212: The initial site sends the key of the target data and the target data to other target sites in the advanced circle.

S213, other target sites store target data.

Other target sites store the target data, and other target sites update the locally recorded metadata, recording the correspondence between the key of the target data and the storage address of the target data.

S214, other target sites create/update replicas in the advanced circle layer according to a preset method in the replica placement strategy configuration module. The execution of S214 is specifically similar to S208. The difference is that S208 is the operation of the target site on the copy in the lowest level circle, while S214 is the operation of other target sites on the copy in the high level circle.

In S215, other target sites notify the initial site that the writing is successful.

In the case that the initial site belongs to multiple high-level circles, the execution of S211 to S215 can be performed in parallel for multiple rounds, each time extending into a higher-level circle. The first execution of S211 to S215 writes the target data in the second level circle, the second execution of S211 to S215 writes the target data in the third level circle, and so on, up to the highest level where the initial site is located The target data has been written in the circle.

For example, the target site determined by the first-level circle-level addressing module of the initial site in S205, the second-level circle-level addressing module of the initial site in S211 continues to search for other target sites in the second-level circle and executes S212 to S215, in parallel, the third-level circle-level addressing module of the initial site continues to search for other target sites in the third-level circle and executes S212 to S215. This allows the target data to be synchronized to multiple high-level circles, ensuring that the client can still access the target data even if the client subsequently moves in a larger range.

If the initial site does not belong to the advanced circle, the initial site can send the target data and the key of the target data to other sites belonging to the advanced circle in the lowest circle to perform S211.

When the data operation request is a read request, Figure 8 shows the flow of processing the data operation request by the edge system.

S301: The client requests the edge gateway to send a site access request to request address information of the site.

Same as S201, the edge gateway selects an initial site for the client.

S302: The edge gateway sends the selected address information of the initial site to the client.

S303: The client may send a data operation request or a service call request (in the case where the client's direct purpose is to call a service running on the site) to the initial site.

S304: The initial site obtains the data operation request, and parses the data operation request through the storage interface processing module.

In the case that the client sends a service invocation request in S303, the service running on the initial site generates a data operation request according to the service invocation request.

The data operation request includes the key of the target data. Optionally, the data operation request may also include circle layer information, and the circle layer information may include a default or indicate a certain level of circle layer. among them,

Default: Instruct the site to address the site where the destination data is located from the lowest level circle. If the site where the destination data is located cannot be determined at the lowest level circle level, the site where the destination data is located will be addressed at the high level circle level. .

Specified_layer: Instructs the site to address the site where the destination data is located only in the circle layer of the specified level.

Generally speaking, when the circle layer information carried in the data operation request when the target data is written (S203) indicates the default, the circle layer information carried in the data operation request when the target data is subsequently read (S303) also indicates default. When the circle layer information carried in the data operation request when the target data is written (S203) indicates the designated circle layer, the circle layer information carried in the data operation request when the target data is subsequently read (S303) also indicates the designated circle layer. Floor.

S305, the lowest-level circle-level addressing module of the initial site, addresses the lowest level circle-level where the initial site is located according to the key of the target data, and determines the target site corresponding to the key of the target data.

S306: The initial site sends the key of the target data to the target site.

After determining the target site, the initial site determines the address information of the target site according to the index table recorded by itself. According to the address information of the target site, the key of the target data is sent to the target site.

S307: The target site determines whether the target data is stored locally.

The target site queries the local metadata according to the key of the target data, and judges whether the target data is stored locally at the target site.

If the destination data is stored at the destination site, execute S3081 and S3082.

S3081: The destination site sends the destination data to the original site.

S3082: The initial site sends the destination data to the client.

In S3081, the target site can also directly send the target data to the client, and the execution of S3082 is omitted.

If the destination data is not stored in the destination site, perform S3083, S3084 and subsequent steps.

S3083, the advanced circle layer addressing module of the destination site, addresses in the advanced circle layer where the destination site is located according to the key of the target data, and determines other target sites corresponding to the key of the target data.

S3084: The target site sends the key of the target data to other target sites.

S309: The other target sites determine whether the target data is stored locally.

Other target sites query local metadata based on the key of the target data to determine whether the target data is stored locally on other target sites.

If the destination data is stored in other destination sites, execute S3101 and S3102.

S3101, other destination sites send the destination data to the original site.

S3102: The initial site sends the target data to the client.

In S3101, other destination sites can also directly send the destination data to the client, omitting the execution of S3102.

If the destination data is not stored in other destination sites, perform S3103, S3104 and subsequent steps.

S3103, the higher level addressing module of other destination sites, addresses the higher level ring layer where the destination site is located according to the key of the target data, and determines the target site where the key of the target data is located in the higher level ring layer.

S3104, other destination sites send the key of the target data to other target sites in the advanced circle.

S307 to S3084 can be repeated multiple times during the process of extending from the low-level circle to the high-level circle. The above shows the first repeated process (S309 to S3104). For example, in S305, the first-level circle-level addressing module of the initial site is addressed in the first-level circle, and in S3083, the second-level circle-level addressing module of the destination site is addressed in the second-level circle. In S3103, the third-level circle layer addressing modules of other destination sites are addressed in the third-level circle layer.

If the initial site of the destination site of the lowest-level circle does not belong to the high-level circle, the destination site of the lowest-level circle can send the key of the target data to other sites in the lowest-level circle that belong to the high-level circle to perform S3083.

In the process of repeating the query of the target data from the lower level to the higher level, if the destination site in any circle finds the target data, it will stop querying the destination site in the higher level circle, and the data can be retrieved Stored in the destination node of the lowest level circle, so that the destination data can be read later. If the destination data still cannot be queried in the highest-level circle, the destination site in the highest-level circle notifies the initial site that the destination data query failed.

S3083 and S3084 (S3103 and S3104) can be executed by the destination site (other destination sites) or the initial site. When some optimized addressing algorithms are used, such as distributed sloppy hash table (DSHT) ), executed by the destination site (other destination sites) can quickly find the destination site in the higher level circle (fewer hops).

If the circle layer information carried in the data operation request is the specified level, the processing flow of the data operation request is as described in S305 to S307 (the lowest level circle layer is replaced with the specified level circle layer). After S307, if the target data is stored in To specify the destination site of the level circle, S3081 and S3082 are executed. If the destination data is not stored in the destination site of the designated tier, the destination site of the designated tier will notify the initial site of the failure of the destination data query.

FIG. 9 provides a computer 400, and each site is implemented by one or more computers 400. In the case where the operating unit in the site is a physical server, the operating unit is the computer 400. When the operating unit in the site is a virtual machine, the operating unit is a virtual machine running on the computer 400.

The computer 400 includes a bus 402, a processor 401, and a storage device 403. The processor 401 and the storage device 403 communicate through a bus 402. The processor 401 may be a central processing unit (CPU). The storage device 403 may include a volatile memory device (volatile memory), such as a random access memory device (RAM), or a non-volatile memory device (non-volatile memory), such as a read-only memory device (read-only memory device). -only memory, ROM), flash storage device, HDD or SSD, etc.

The storage device 403 stores executable instructions, and the processor 401 executes the executable instructions to execute the storage interface processing module, addressing domain configuration module, addressing module, data storage module, and service (optional) in FIG. . The storage device 403 may also include executable instructions required to run an operating system (operation system, OS). The OS can be LINUX ^TM , UNIX ^TM , WINDOWS ^TM and so on.

The descriptions of the processes corresponding to each of the above figures have their respective focuses. For parts that are not described in detail in a certain process, please refer to related descriptions of other processes.

In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, it can be implemented in the form of a computer program product in whole or in part. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the processes or functions according to the embodiments of the present invention are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instruction may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instruction may be transmitted from a website, a computer, a server, or a data center through a cable (Such as coaxial cable, optical fiber, digital tenant line or wireless (such as infrared, wireless, microwave, etc.) transmission to another website site, computer, server or data center. The computer-readable storage medium can be accessed by a computer Any usable medium in the data storage device or a data storage device including one or more usable media integrated servers, data centers, etc. The usable medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), Or semiconductor media (such as SSD) and so on.

Claims

An edge system, characterized in that the edge system includes a plurality of first-level circles and second-level circles, and the geographic area represented by the second-level circles includes the geographic regions represented by each first-level circle. area;

Each first-level circle includes multiple sites, each site is used to process data operation requests sent by the client, and an initial site among the multiple sites is also included in the second-level circle;

The initial site is used to process the data operation request in the first-level circle and the second circle that include the initial site.
The edge system of claim 1, wherein the data operation request is a write request;

The initial site is configured to address the target site corresponding to the target data carried in the data operation request within the first-level circle including the initial site, and send the target data to the first site that includes the initial site A target site of the first-level circle; address the target site corresponding to the target data in the second-level circle, and send the target data to the target site of the second-level circle;

The target site of the first-level circle layer including the initial site is used to store the target data;

The target site of the second level circle layer is used to store the target data.
The edge system according to claim 2, wherein the target site of the first-level circle layer including the initial site is also used to create or create on other sites of the first-level circle layer including the initial site. Update the copy of the target data.
The edge system according to claim 2 or 3, wherein the target site of the second-level circle is also used to create or update the target data on other sites of the second-level circle. Copy.
The edge system according to claim 1, wherein the data operation request is a read request;

The initial site is configured to address the target site corresponding to the key of the target data carried in the data operation request within the first-level circle layer including the initial site;

The target site including the first-level circle layer of the initial site is used to determine whether the target data is stored locally in the target site of the first-level circle layer including the initial site, and if the target data is stored in the first-level circle layer including the initial site, The target site of the first level circle layer of the initial site is locally sent, and the target data is sent to the client. If the target data is not stored locally in the target site of the first level circle layer including the initial site, Address the target site corresponding to the key of the target data in the second-level circle;

The target site of the second-level circle is used to determine that the target data is stored locally at the target site of the second-level circle, and send the target data to the client.
The edge system of claim 5, wherein the target site of the second level circle layer is further used to send the target data to the target site of the first level circle layer including the initial site.
A method for processing a data operation request, wherein the method for processing a data operation request is applied to an edge system, and the edge system includes a plurality of first-level circle layers and second-level circle layers. The geographic area represented by the circle layer includes the geographic area represented by each first-level circle layer, each first-level circle layer includes multiple sites, and each site is used to process data operation requests sent by the client. The multiple sites An initial site is included in the second-level circle layer at the same time;

Addressing, by the initial site, the target site corresponding to the target data carried in the data operation request within the first-level circle layer including the initial site;

Sending, by the initial site, the target data to a target site in the first-level circle layer including the initial site;

The target site including the first-level circle layer of the initial site stores the target data;

The initial site addresses the target site corresponding to the target data in the second-level circle, and sends the target data to the target site of the second-level circle;

The target site of the second-level circle layer stores the target data.
The method of claim 7, further comprising:

The target site including the first-level circle of the initial site creates or updates a copy of the target data on other sites including the first-level circle of the initial site.
The edge system according to claim 7 or 8, further comprising:

The target site of the second-level circle creates or updates a copy of the target data on other sites of the second-level circle.
A method for processing a data operation request, characterized in that the method for processing a data operation request is applied to an edge system, and the edge system includes a plurality of first-level circles and second-level circles, and the second-level The geographical area represented by the circle layer includes the geographical area represented by each first-level circle layer. Each first-level circle layer includes multiple sites. Each site is used to process data operation requests sent by the client. The stations include an initial station and are also included in the second-level circle layer;

The initial site addresses the target site corresponding to the key of the target data carried in the data operation request within the first-level circle layer including the initial site;

The target site of the first-level circle including the initial site determines whether the target data is stored locally at the target site of the first-level circle including the initial site;

If the target data is stored locally at a target site including the first-level circle layer of the initial site, sending the target data to the client;

If the target data is not stored locally in the target site of the first level circle layer including the initial site, address the target site corresponding to the key of the target data in the second level circle layer;

The target site of the second-level circle determines that the target data is stored locally at the target site of the second-level circle, and sends the target data to the client.
The method of claim 10, further comprising:

The target station of the second level circle layer sends the target data to the target station of the first level circle layer including the initial station.
A non-transitory readable storage medium, characterized in that, when instructions stored in the non-transitory readable storage medium are executed by at least one site in an edge system, the edge system executes claims 7 to 11 Any of the methods described.
A computer program product, wherein when the instructions contained in the computer program product are executed by an edge system, the edge system executes the method according to any one of claims 7 to 11.